KR102004513B1 - Adaptive Phase-Distortionless Magnitude Response Equalization (MRE) for beamforming applications - Google Patents

Abstract

근사들로 계산될 수 있다. 그 후, 형 필터는 프로세싱된 신호들에 기초하여 계산된 필터를 사용하여 그의 원래의 시간 도메인 형태의 입력 센서 신호들을 등화시킬 수 있다. The time domain impulse response filter may be used to equalize signals in the time domain to avoid errors and artifacts introduced by domain transforms such as IFFTs. The disclosed time domain impulse response filter is based on the magnitude response of the individual signals. The magnitude responses of each signal may be used in the frequency domain or in other techniques, such as auto-regression analysis and mathematical signal approximation algorithms,

Can be calculated as approximations. The type filter can then equalize the input sensor signals in its original time domain form using the calculated filter based on the processed signals.

Description

Adaptive Phase-Distortionless Magnitude Response Equalization (MRE) for beamforming applications

[0001] This disclosure relates to magnitude response equalization in multi-sensor systems. More particularly, portions of the present disclosure relate to a magnitude response equalization of signals from multiple microphone systems using adaptive filtering in the time domain.

Systems comprising multiple microphones may be used to detect directional sound by using beam forming techniques in which signals from at least two microphones are compared to observe phase shifts and magnitude differences . The processing of signals from two different microphones capturing the same sounds requires equalization because the physical characteristics and magnitude responses may vary between the microphones. These variations may exist between microphones of the same make and model due to minor manufacturing variations. Variations may also be caused by a number of other factors such as microphone boots, tube length differences and other variations. Applications such as beamforming assume that the differences in the signals measured in the respective microphones are due only to environmental and spatial differences, not the difference in the manner in which the signals are measured. Complicating the processing of signals. Thus, signal processing in multi-microphone systems attempts to equalize the raw signals to improve the accuracy of the signal processing calculations.

[0003] One conventional technique for equalization is off-line calibration during system production. This technique requires the manufacture of microphones with extremely low tolerance errors, which increases the cost and sensitivity of the microphones. Another prior art for equalization is self-calibration. On-line self-calibration using gain or magnitude response techniques involves calculating propagation loss and phase matching. On-line self-calibration using frequency response techniques requires knowledge of the location of the control stimulus.

[0004] On-line self-calibrations using magnitude response techniques generally include time domain signals (eg, two signals from two separate microphones) for each microphone in the frequency domain And then calculating an equalization ratio based on the first and second signals over the frequency range. The equalization ratio is then applied to the frequency domain of the second signal in an attempt to match it to the first microphone. The adjusted second signal is then converted back to the time domain and further processing such as beamforming calculations may be performed in conjunction with the first and second signals. This technique reduces errors introduced by variations in two microphones but introduces additional errors in equalization computations.

[0005] Manipulating the frequency domain of the second signal using an equalization ratio calculated over all frequencies and then converting back to the time domain introduces errors into the calculations. The magnitude response of the microphones varies over frequencies so that the calculated equalization ratio only approximates the magnitude difference of only two signals and does not take into account the fluctuating magnitude response of different microphones at different frequencies. Also, when converting the adjusted second signal back from the frequency domain to the time domain, the signal generated by the Inverse Fast Fourier Transform (I-FFT) inherently introduces errors due to the mathematical limitations of the I-FFTs. This prior art technique is illustrated in Figure 1 where a frequency domain signal for x ₂ [n] is taken from node 101 after conversion to frequency domain at block 105 and a frequency domain response Lt; RTI ID = 0.0 > 102 < / RTI > The equalized frequency response of x ₂ [n] is then transformed in the I-FFT block 104.

[0006] The disadvantages mentioned here are merely exemplary and are included to simply emphasize the need for improved electrical components for multiple microphone systems, particularly those used in consumer-level devices such as mobile phones . The embodiments described herein solve certain drawbacks, but do not necessarily solve each and every disadvantage described herein or known in the art.

Magnitude response equalization of multiple sensor systems can be improved by using a time domain impulse response filter based on the magnitude responses of individual signals to equalize the magnitude response of multiple microphones over a desired frequency spectrum. Conventional techniques equalize signals in the frequency domain, which generates errors and artifacts that propagate into the time domain representation of the equalized signal as the equalized signal is transformed from the frequency domain to the time domain. The methods and apparatus described herein reduce or eliminate signal errors introduced by conventional frequency domain equalization techniques by creating a time domain impulse response filter that equalizes signals in the time domain. Thus, errors and artifacts introduced by domain transforms such as I-FFT are avoided. In addition, the signal processing is limited to reduce or prevent the introduction of phase differences between the input signals.

근사들과 같은 자동-회귀 분석 및 수학적 신호 근사 알고리즘들로 계산될 수 있다. 제 2 마이크로폰을 제 1 마이크로폰과 등화시키기 위해 시간 도메인에서 시스템의 마이크로폰들의 매그니튜드 응답에 기초한 시간 도메인 임펄스 응답 필터를 적용하는 것은 제 2 신호의 등화가 주파수 도메인에서 행해지는 종래 기술의 시스템들에서 도입된 에러를 회피한다. [0008] In some embodiments, the time domain impulse response filter is based on the magnitude responses of individual signals and is used to equalize the magnitude responses of multiple microphones over a desired frequency spectrum. The magnitude responses of each signal may be transmitted in the frequency domain or by other techniques,

Regression analysis such as approximations and mathematical signal approximation algorithms. Applying a time domain impulse response filter based on the magnitude response of the microphones of the system in the time domain to equalize the second microphone with the first microphone is advantageous in that the equalization of the second signal is introduced in prior art systems in the frequency domain Avoid errors.

[0009] According to one embodiment, a method includes receiving, by a processor coupled to a plurality of sensors, at least a first input signal and a second input signal in a time domain from a plurality of sensors; Converting, by the processor, the first input signal and the second input signal from a time domain to a frequency domain input signal; Estimating, by the processor, a magnitude response difference between the first input signal and the second input signal based at least in part on the frequency domain input signal; Converting, by the processor, the magnitude response difference to a time domain impulse response; Limiting the time domain impulse response by the processor to have a linear phase response; And / or by the processor, filtering at least one of the first input signal and the second input signal based at least in part on the limited time domain impulse response.

[0010] In some embodiments, the filtering may comprise equalizing the magnitude response between the first input signal and the second input signal received from the plurality of sensors; Estimating the magnitude response difference comprises calculating filter coefficients for the adaptive filter; The step of limiting includes limiting filter coefficients to even symmetric and odd lengths, and the step of filtering may comprise applying an adaptive filter having calculated and limited filter coefficients.

[0011] In some embodiments, the method further includes repeating the steps of receiving, estimating, converting, limiting and filtering to provide adaptive equalization of the received input signals; Delaying at least one of the first and second unfiltered input signals based on a limited time domain impulse response to compensate for the delay introduced by the filtering, The filtered second input signal may be further filtered for spatial recognition; And / or the first input signal and the filtered second input signal may be further filtered for beamforming.

[0012] According to another embodiment, an apparatus includes: a first input node configured to receive a first input signal; A second input node configured to receive a second input signal; And / or a controller coupled to the first input node and coupled to the second input node. The controller comprises: receiving a first input signal and a second input signal in a time domain; Converting a first input signal and a second input signal from a time domain to a frequency domain input signal; Estimating a magnitude response difference between the first input signal and the second input signal based at least in part on the frequency domain input signal; Converting the magnitude response difference to a time domain impulse response; Limiting the time domain impulse response to have a linear phase response; And / or filtering at least one of the first input signal and the second input signal based at least in part on the limited time domain impulse response.

[0013] In some embodiments, the controller may perform the filtering by equalizing the magnitude response between the first input signal and the second input signal received from the plurality of sensors; And / or estimating the magnitude response difference by calculating filter coefficients for the adaptive filter, wherein the limiting step comprises limiting filter coefficients to even symmetric and odd lengths, Comprises applying an adaptive filter having calculated and constrained filter coefficients.

[0014] In some embodiments, the controller is further configured to repeat the receiving, estimating, converting, limiting and filtering steps to provide adaptive equalization of the received input signals ; And to delay at least one of the unfiltered first input signal and the second input signal based on a limited time domain impulse response to compensate for the delay introduced by the filtering and / or filtering.

근사를 사용하여 계산됨 ― ; 및 프로세서에 의해, 선형 위상 응답을 갖도록 시간 도메인 임펄스 응답을 제한하는 단계; 및/또는 프로세서에 의해, 제한된 시간 도메인 임펄스 응답에 적어도 부분적으로 기초하여 제 1 입력 신호 및 제 2 입력 신호 중 적어도 하나를 필터링하는 단계를 포함할 수 있다. [0015] According to another embodiment, a method includes receiving, by a processor from a plurality of sensors, at least a first input signal and a second input signal in a time domain; Computing, by the processor, auto-regressive (AR) model parameters of the input signals by using linear prediction analysis; Computing, by the processor, auto-regressive moving average (ARMA) model parameters corresponding to the magnitude response differences between the two input signals; Computing a time domain impulse response by a processor corresponding to a magnitude response difference between a first input signal and a second input signal, wherein the magnitude response difference is at least equal to the autoregressive model parameters and the autoregressive moving average model parameters On a partial basis

Calculated using approximation; And limiting, by the processor, the time domain impulse response to have a linear phase response; And / or by the processor, filtering at least one of the first input signal and the second input signal based at least in part on the limited time domain impulse response.

[0016] In some embodiments, applying the linear prediction analysis may comprise generating linear prediction coefficients; And / or the first input signal and the second input signals may comprise audio information.

근사를 사용하여 계산됨 ― ; 및 선형 위상 응답을 갖도록 시간 도메인 임펄스 응답을 제한하는 단계; 및/또는 제한된 시간 도메인 임펄스 응답에 적어도 부분적으로 기초하여 제 1 입력 신호 및 제 2 입력 신호 중 적어도 하나를 필터링하는 단계를 포함하는 단계들을 수행하도록 구성될 수 있다. [0017] In yet another embodiment, an apparatus includes a first input node configured to receive a first audio signal; A second input node configured to receive a second audio signal; And / or a controller coupled to the first input node and coupled to the second input node. The controller comprises: receiving a first input signal and a second input signal in a time domain; Computing, by the processor, auto-regressive (AR) model parameters of the input signals using linear prediction analysis; Computing, by the processor, auto-regressive moving average (ARMA) model parameters corresponding to the magnitude response differences between the two input signals; Computing a time domain impulse response by a processor corresponding to a magnitude response difference between a first input signal and a second input signal, wherein the magnitude response difference is at least equal to the autoregressive model parameters and the auto- On a partial basis

Calculated using approximation; And limiting the time domain impulse response to have a linear phase response; And / or filtering at least one of the first input signal and the second input signal based at least in part on the limited time domain impulse response.

[0018] In certain embodiments, the controller may be configured to apply linear prediction analysis by generating linear prediction coefficients; The first input signal and the second input signals may comprise audio information; And / or the audio information may be audio information received from the first microphone and the second microphone.

[0019] The foregoing has outlined rather broadly certain features and technical advantages of embodiments of the invention in order that the detailed description that follows may be better understood. Additional features and advantages forming the gist of the claims of the present invention will be described below. It should be appreciated by those skilled in the art that the disclosed concepts and specific embodiments may be readily utilized as a basis for modifying or designing other structures for carrying out the same or similar purposes. It should also be appreciated by those skilled in the art that such equivalent constructions do not depart from the spirit and scope of the invention as set forth in the appended claims. Additional features will be better understood from the following description when considered in connection with the accompanying drawings. It is to be expressly understood, however, that each of the figures is provided for purposes of illustration and description only, and is not intended to limit the invention.

[0020] For a more complete understanding of the disclosed systems and methods, reference is now made to the following description taken in conjunction with the accompanying drawings.
[0021] FIG. 1 is an exemplary block diagram of a system for equalizing a second signal for a first signal in the frequency domain in accordance with the prior art.
[0022] FIG. 2 is an exemplary block diagram of an adaptive filter for equalizing a second signal for a first signal in the time domain, where the adaptive filter is based on the magnitude response of the first and second signals.
[0023] FIG. 3 illustrates exemplary steps for equalizing a second signal for a first signal in the time domain with an adaptive filter based on the magnitude response of the first and second signals, in accordance with an embodiment of the present disclosure; Fig.
[0024] FIG. 4 is a block diagram of an adaptive filter, based on the magnitude response of first and second signals computed in the frequency domain, to equalize a second signal to a first signal in the time domain, according to one embodiment of the present disclosure Lt; RTI ID = 0.0 > flowchart < / RTI >
[0025] FIG. 5 illustrates an example of an adaptive filter that is based on the magnitude response of first and second signals computed in the frequency domain, in accordance with one embodiment of the present disclosure, to equalize the second signal to the first signal in the time domain ≪ / RTI > FIG.
[0026] FIG. 6A is an exemplary graph illustrating the magnitude response of two microphones without equalization, in accordance with an embodiment of the present disclosure.
[0027] FIG. 6B is an exemplary graph illustrating the magnitude response of two microphones after applying the magnitude response equalization technique in accordance with an embodiment of the present invention.
[0028] FIG. 7 is a graphical representation of an adaptive filter based on the magnitude response of first and second signals computed in the time domain using auto-regressive modeling, in accordance with one embodiment of the present disclosure, ≪ / RTI > is an exemplary flow chart of exemplary steps for equalizing a second signal with respect to a second signal.
[0029] FIG. 8 is an exemplary block diagram of an adaptive filter for equalizing a second signal for a first signal in the time domain, in accordance with one embodiment of the present disclosure, wherein the adaptive filter includes auto- Based on the magnitude response of the first and second signals calculated in the time domain.

[0030] An example of mismatches and variations in the magnitude response of different microphones in a multiple microphone system, which may be addressed by embodiments of the present disclosure, is shown in FIG. 6A. The graph of FIG. 6A illustrates the magnitude response of two microphones to a control signal at lines 602 and 604. FIG. For example, due to mike mismatches that may occur during manufacture, the microphones respond differently to the stimulation of each frequency. Equalizing the response of one microphone to the response of another microphone can improve the processing of the audio captured by the microphones, such as the user's voice. In some embodiments, a time domain impulse response filter may be applied during equalization of the signals.

및

)은 제한된 시간 도메인 임펄스 응답 필터(204)를 계산하도록 프로세싱 블록(203)에서 사용된다. 그 후, 제한된 시간 도메인 임펄스 응답 필터(h[n])는, 제 2 센서로부터의 제 2 신호(x₂[n])에 대해 제 1 센서로부터의 제 1 신호(x₁[n])를 등화시키도록 필터(204)에 의해 시간 도메인 입력 신호들 중 하나에 적용된다. 일 실시예에서, 지연 블록(205)은 필터(204)에 의해 도입된 지연을 보상하기 위해 프로세싱 블록(201)의 것과 같은 매그니튜드 응답 계산 후에 삽입될 수 있다. [0031] Referring now to FIG. 2, one technique for equalizing the magnitude response of two microphones is shown. Figure 2 illustrates such an exemplary system 200 for implementing magnitude response equalization with an adaptive filter in accordance with one embodiment of the present disclosure. The input signals x ₁ [n] and x ₂ [n], such as time domain audio signals from the first and second microphones, are received at the input nodes 211 and 212 of the signal processing system 200. The signals x ₁ [n] and x ₂ [n] are provided to the processing blocks 201 and 202, which compute the magnitude response for each time domain signal. Thereafter, the calculated magnitude response (

And

Is used in processing block 203 to compute a limited time domain impulse response filter 204. [ The limited time domain impulse response filter h [n] then subtracts the first signal x ₁ [n] from the first sensor for the second signal x ₂ [n] from the second sensor Is applied to one of the time domain input signals by the filter 204 to equalize. In one embodiment, the delay block 205 may be inserted after a magnitude response calculation such as that of the processing block 201 to compensate for the delay introduced by the filter 204.

Although it has been described in certain embodiments that the signals x ₁ [n] and x ₂ [n] are microphone signals such as those received from a digital microelectromechanical systems (MEMS) microphone, And can be processed with the systems and methods described herein. The input signals x ₁ [n] and x ₂ [n] may be digital signals of a time domain representation. The input signals x ₁ [n] and x ₂ [n] may be received directly from memory, buffers, or analog-to-digital converters (ADCs) coupled to sensors or microphones.

[0033] The magnitude response equalization of FIG. 2 is advantageous because the equalization of the second signal for the first signal is performed using a time domain filter based on the magnitude response of both microphones, It is possible to provide better matching. This matching reduces the error introduced in conventional systems in which the equalization of the second microphone is typically performed in the frequency domain and then the equalized second microphone signal is converted from the frequency domain to the time domain. 3 is an exemplary signal processing flow for matching a magnitude response in the time domain in accordance with one embodiment of the present disclosure. The time domain input signals x ₁ [n] and x ₂ [n] are received at blocks 301 and 302, respectively, from the input nodes. The magnitude response for each of the signals x ₁ [n] and x ₂ [n] is computed in blocks 303 and 304, respectively. The magnitude response of each signal can be estimated in the time domain or the frequency domain or a combination of both. After calculating the magnitude response at blocks 303 and 304, a time domain impulse response based on the calculated magnitude responses is calculated at block 305. [ Since the time domain impulse response may include some phase distortion, the time domain impulse response may be limited at block 306. [ Then, for the block at 307, one of the limited time domain impulse response of the input signal, for example, is applied to the x ₂ [n], filter the signal, the microphone receives signals (x ₁ [n]) And equalizes the microphone response of the microphone received signal x ₂ [n].

Limiting the time domain impulse response minimizes or zeroes the introduction of phase distortion for the signal (x ₁ [n] or x ₂ [n]). Beamforming and other signal processing techniques calculate parameters based on the time difference of arrival of signals received at the microphones. The time-of-arrival information can be changed when the phase information of the microphone signals is distorted by the signal processing techniques. By limiting the impulse response, the phase distortion can be reduced or eliminated, so that there is no significant effect on subsequent signal processing. For example, beamforming depends on the phase difference information between the microphone signals (x ₁ [n] and x ₂ [n]) to form a beam or a null in a particular direction. Limiting the response at block 307 allows beamforming or null forming to operate with reduced error.

The signals used to generate the magnitude response equalization filter, for example, the filter of block 307 of FIG. 3 and the filter h [n] of the processing block 204 of FIG. 2, . In some embodiments, the signal may be processed to produce a uniform magnitude, e.g., white noise, over a desired frequency range. However, the magnitude response equalization can be applied to any input signal at any time and does not require a control signal with a uniform magnitude response over the frequency ranges.

[0036] In some embodiments, the magnitude response equalization applied in generating the adaptive filter may be calculated using frequency domain representations of the signals. 4 is an exemplary signal processing flow for matching a magnitude response in the time domain in accordance with one embodiment of the present disclosure wherein the adaptive filter is based on the magnitude response of the signals in the frequency domain in accordance with one embodiment of the present disclosure . In the exemplary flow of FIG. 4, at least two signals from at least two distinct sensors are received in blocks 401 and 402. In some embodiments, two signals x ₁ [n] and x ₂ [n] are received in the time domain from the first and second sensors, respectively. The input signals x ₁ [n] and x ₂ [n] are then converted to the frequency domain in blocks 403 and 404, respectively. The frequency domain representations of the signals x ₁ [n] and x ₂ [n] are shown as frequency domain representations X ₁ (z) and X ₂ (z), respectively, in blocks 403 and 405 , Other frequency domain representations may be used in some embodiments. The magnitude response difference between the frequency domain representations X ₁ (z) and X ₂ (z) is calculated at block 405. The magnitude response difference includes coefficients representing the difference in magnitude response for sensor 1 and sensor 2 at various frequencies. The magnitude response difference is then converted, at block 406, to a time domain impulse response filter h [n]. In some embodiments, the filter h [n] is an adaptive filter. In some embodiments, the time domain impulse response filter h [n] is limited to having a linear phase at block 407 to prevent phase distortions when applying a filter h [n] to the input signal do. The filter h [n] is then applied at block 408 to one of the input signals, e.g., signal x ₂ [n].

[0037] The adaptive filter may be calculated after frequency domain conversion of the signals as shown in the system of FIG. The system of Figure 5 receives, at nodes 501 and 502, input signals x ₁ [n] and x ₂ [n] from first and second sensors, such as two microphones. Nodes 501 and 502 are coupled to respective processing blocks 503 and 504 where time domain signals x ₁ [n] and x ₂ [n] are buffered, windowed ) And / or overlapping. Processing blocks 503 and 504 are coupled to respective Fast Fourier transform (FFT) processing blocks 505 and 506, wherein input signals x ₁ [n] and x ₂ [n] . The FFT processing block 505 is coupled to the magnitude smoothing blocks 507 and 509 and the FFT processing block 506 is coupled to the magnitude smoothing blocks 508 and 510. The magnitude smoothing blocks may be computed using the following methods: average squared displacements (shown in processing blocks 509 and 510), cepstrum method, running average filtering, Savitzky-Golay smoothing or other smoothing One of the algorithms can be used to estimate the magnitude spectral density (MSD). The magnitude smoothing blocks 507-510 may perform magnitude smoothing in software or in hardware. Magnitude smoothing in hardware components may be accomplished using, for example, a low-pass or band-pass filter.

및

)을 각각 생성한다. I-FFT 블록(511)은 적응형 필터(514)에 커플링되는 에러 신호 프로세싱 블록(513)에 커플링된다. 적응형 필터(514)는 또한

를 수신하기 위해 I-FFT 프로세싱 블록(512)에 커플링된다. 적응형 필터(514)는 필터(h[n])에 대한 FIR 계수들을 생성하고, 필터(h[n])가 에러 신호 프로세싱 블록(513)에 대한 입력인 피드백 루프를 생성하도록 에러 신호 프로세싱 블록(513)에 추가로 커플링될 수 있다. 적응형 필터(514)에 대한 에러 신호 피드백은

및

의 컨버전스(convergence)를 획득하도록 적응형 필터의 필터(h[n])에 대한 FIR 계수들을 정제한다. 동일한 계수들이 시간 도메인 신호들(x₁[n] 및 x₂[n]) 중 하나에 필터를 적용하는 적응형 필터(515)에 의해 적용될 수 있다. [0038] After such processing in the frequency domain, the signals may be converted back to the time domain and used to generate coefficients for the adaptive filter blocks 514 and 515. Thus, the magnitude smoothing blocks 507 and 509 are coupled to an I-FFT (Inverse-Fast Fourier Transform) block 511 and the magnitude smoothing blocks 508 and 510 are coupled to an I-FFT block 512 do. The I-FFT blocks 511 and 512 are used to generate signals (e. G., Signals) that are time domain representations of the smoothed magnitude spectra of the microphone signals x ₁ [n] and x ₂ [

And

Respectively. The I-FFT block 511 is coupled to an error signal processing block 513 coupled to the adaptive filter 514. The adaptive filter 514 also

FFT processing block 512 to receive the I-FFT processing block 512. The I- The adaptive filter 514 generates FIR coefficients for the filter h [n] and outputs the error signal to the error signal processing block 514 to generate a feedback loop, which is the input to the error signal processing block 513, Lt; RTI ID = 0.0 > 513 < / RTI > The error signal feedback for the adaptive filter 514 is

And

(H [n]) of the adaptive filter to obtain the convergence of the FIR coefficients of the adaptive filter. The same coefficients may be applied by an adaptive filter 515 that applies a filter to one of the time domain signals x ₁ [n] and x ₂ [n].

)을 부과하여서,

가 지연 블록(518)의 출력이 되며,

는 에러 신호가 에러 신호 프로세싱 블록(513)에서 계산될 때, 적응형 필터(514)를 통과한

와 동기화된다. [0039] In some embodiments, the I-FFT processing block 511 is further coupled to a delay block 518 between the I-FFT block 511 and the error signal processing block 513, , For example, a simple delay (e < RTI ID = 0.0 >

),

Is the output of delay block 518,

When the error signal is computed in the error signal processing block 513, passes through the adaptive filter 514

≪ / RTI >

[0040] Referring back to processing blocks 503 and 504, blocks 503 and 504 buffer, overlap, and / or window the input signals, The signals can be processed by converting:

Where n [n] is the windowing function, xi [n, m] is the buffered and superimposed input signal corresponding to the m-th superframe, N is the FFT size that can be changed through tunable parameters, l Is a frequency bin index. The overlap can be fixed at 50% and a Kaiser-Bessel induction window can be used in this analysis stage. The performance of magnitude responsive equalization systems and methods is not limited by the window function as a whole. In some embodiments, a window other than a rectangular window may be applied.

[0041] Referring now to processing blocks 507, 508, 509, and 510, the magnitude spectrum can be computed from the complex frequency spectrum and smoothed using a first order exponential averaging filter based on the following equation:

,

,

Where alpha is a smoothing parameter that can be changed by the user or an algorithm running on the processor.

[0042] The smoothed magnitude spectrum may then be transformed into the time domain using inverse Fourier transforms in blocks 511 and 512 based on the following equation:

)는 입력 신호(x_i[n])가 컬러링 필터(g_i[n])에 의해 화이트 노이즈 신호를 필터링함으로써 획득되는 것으로 가정함으로써 해석될 수 있다. The output signal (

) Can be interpreted by assuming that the input signal x _i [n] is obtained by filtering the white noise signal by the coloring filter g _i [n].

[0044] For a wide sense stationary system (WSS)

이고,

ego,

는 입력 신호(x_i[n])의 파워 스펙트럼 밀도이고,

는 컬러링 필터의 주파수 응답이고,

는 여기 화이트 노이즈 신호의 주파수 응답이다. WSS 가정에 있어서, 출력 신호(

)는 다음과 같이 작성될 수 있다:here,

Is the power spectral density of the input signal x _i [n]

Is the frequency response of the coloring filter,

Is the frequency response of the white noise signal here. In the WSS hypothesis, the output signal (

) Can be written as:

.

.

)는 컬러링 필터(g_i[n])의 매그니튜드 응답 정보만을 포함한다. MRE 시스템 및 방법들의 목적은 컬러링 필터들의 매그니튜드 응답을 추정하고 컬러링 필터들 중 하나의 매그니튜드 응답을 다른 것에 대한 매그니튜드 응답과 매칭시키는 등화 필터를 설계하는 것이다. 이 등화 필터의 매그니튜드 응답은 다음과 같을 수 있다:[0045] Therefore, the signal (

) Contains only the magnitude response information of the coloring filter g _i [n]. The objective of the MRE system and methods is to design an equalization filter that estimates the magnitude response of the coloring filters and matches the magnitude response of one of the coloring filters with the magnitude response of the other. The magnitude response of this equalization filter may be:

[0046] Magnitude difference compensation can be implemented in the frequency domain by multiplying the complex spectrum of one of the microphone signals by a real gain function, as was done in the prior art. However, such scaling in the frequency domain may introduce artifacts into the synthesized time domain signal. Embodiments described herein perform a time domain filter, e. G., Equalization through a FIR filter, instead. The filter coefficients are estimated through an adaptive filter operating on a time domain representation of the smoothed magnitude spectrum of the microphone signals. In some embodiments, the magnitude response equalization block may equalize only the magnitude response differences. Therefore, the coefficients can be updated in such a way that the phase response of the filter is limited to be linear. This linear phase response can be replaced by the introduction of a simple delay in the equalized output. The reference to the adaptive filter is defined as:

,

,

Where L is the number of filter coefficients that can be tuned through the input parameters. The error signal is then given by the following equation.

,

,

는, 매그니튜드 스펙트럼이 필터 계수들에 의해 필터링되는 기준 신호에 의해 매칭되어야 하는 신호의 지연된 버전이다. 제한되지 않은 적응형 필터에 대한 필터 계수들은 다음과 같이 NLMS(normalized least mean squares) 리커시브 업데이트 등식을 사용하여 획득될 수 있다 :here,

Is a delayed version of the signal whose magnitude spectrum should be matched by the reference signal filtered by the filter coefficients. The filter coefficients for the unconstrained adaptive filter can be obtained using a normalized least mean squares (NLMS) recursive update equation as follows:

,

,

는 0에 의한 나눗셈을 방지하기 위한 작은 정규화 요소이다. 선형 위상 제한 적응형 필터 업데이트 등식은 선형 위상 FIR 필터의 계수 대칭 속성들의 이용을 통해 위의 등식을 수정함으로써 획득될 수 있다. FIR 필터의 이동 평균 형태는 다음 등식에 의해 주어질 수 있다:here,

Is a small normalization factor to prevent division by zero. The linear phase limited adaptive filter update equation can be obtained by modifying the above equation through the use of the coefficient symmetry properties of a linear phase FIR filter. The moving average form of the FIR filter can be given by the following equation:

[0047] For a Type I linear phase FIR system, the coefficients may be limited to be even symmetric and odd lengths as defined in the following equation:

The delay introduced by this filter may be (L-I) / 2 samples. The output of this filter can be defined as follows

.

.

[0048] Thus, by reordering the reference buffer, the linear phase FIR filter coefficients can be estimated using the standard NLMS update equation. In particular, the reference vector and the coefficient vector may be reduced as follows:

.

.

[0049] The number of coefficients inherent in a type 1 linear phase filter may be ((L-1) / 2 + 1). In some embodiments, only these unique coefficients can be estimated. The NLMS update equation for a linear phase limited FIR filter may be modified as shown in the following equation

.

.

)의 자동-상관은 다음 등식에서 도시된 바와 같이 입력 신호(x_i[n])와 동일할 수 있다:The delay (lambda) may be set for (L-1) / 2 samples to derive an error signal. The adaptive rate may then be tuned through a tunable parameter, selected by the user, or determined by the processor. signal(

) May be equal to the input signal x _i [n] as shown in the following equation:

,

,

)에 기초하여 구현되는 적응형 필터의 컨버전스 속성들이 원래의 입력 신호들(x_i[n])의 자동-상관 속성들에 의해 관리될 수 있음을 의미한다. Where p is the auto-correlation lag index. This relationship is based on signals (

) Can be managed by the auto-correlation attributes of the original input signals x _i [n].

[0051] When the equalization filter coefficients are estimated from the time domain equivalent to the magnitude spectrum, the filter can be applied separately to the raw input signal x ₂ [n]. In particular, the equalized output can be defined by the following equation

.

.

The non-equalized input can be delayed to compensate for the delay introduced by the linear phase FIR filter given by the following equation:

.

.

[0053] The output of the delay block 518 may be y ₁ [n] and the output of the adaptive filter 514 may be y ₂ [n]. The signal (y ₁ [n] and y ₂ [n]) can be further filtered for the beam-forming applications (e.g., beam-forming or spatial filter). For example, the beamforming at y ₁ [n] and y ₂ [n] may include filtering the signals x ₁ [n] and x ₂ [n]. The signal s (x ₁ [n] and x ₂ [n]) is for filtering at least one of the signals to change the phase or magnitude (x ₁ [n] and x ₂ [n]) of the signal (x ₁ [ n] and x2 [n]). In some embodiments, the beamforming filtering using y ₁ [n] and y ₂ [n] may be performed in a spatial-temporal manner, for example, between sensors producing signals x ₁ [n] and x ₂ [n] Can be used to detect the position of the signal source by calculating the magnitude and phase shift differences between y ₁ [n] and y ₂ [n] that are at least partially caused by the relationship.

[0054] FIG. 6a illustrates a spectral plot of two input sensors labeled Mic1 and Mic2. This drawing highlights the problems solved by the systems and methods disclosed herein. Signal comparisons between the data of two sensors with different spectral responses, such as Micl and Mic2 in Figure 6a, must be equalized to perform further processing on the signal, such as beamforming. 6B illustrates a spectral plot of the raw data of Mic1 that is plotted with a spectral plot of the filtered raw data from Mic2 where the Mic2 raw data is filtered using one embodiment of the systems and methods described herein . As shown in FIG. 6B, the filtered Mic2 signal 606 is equalized across the associated frequency spectrum to match the magnitude response of the Mic1 signal 608. The comparison signal analysis between Mic1 and Mic2 may be performed by embodiments herein by eliminating signal processing errors that would normally have been caused by intrinsic or environmental differences of the first and second microphones that produce the signals Mic1 and Mic2 Can be strengthened.

)이 변하기 때문에, 적응형 인에이블 입력 신호 제어 신호가 참일 때만 업데이트될 수 있다. [0055] In some circumstances, the input signal is composed of noise and voice, and the relative magnitude spectrum of voice and noise may be very different. In such scenarios, always matching the magnitude spectrum can cause undesirable results. Accordingly, some embodiments further include an adaptive enable input signal that controls time instances during which smoothed magnitude spectrum estimation is enabled. The adaptive filter is used only when the smoothed magnitude spectrum estimation is enabled,

Is changed, it can be updated only when the adaptive enable input signal control signal is true.

근사를 사용하여 발견된 신호들의 매그니튜드 응답에 기초할 수 있다. 도 7은 입력 신호들(x₁[n] 및 x2[n])이 블록들(701 및 702)에서 각각 수신되는 본 발명의 방법들의 실시예를 예시한다. 블록(703)은 신호(x₁[n])의 AR(Auto-Regressive) 모델 파라미터들의 추정을 계산하고, 블록(704)은 신호(x2[n])의 AR 모델 파라미터들의 추정을 계산한다. 다음으로, ARMA(Auto-Regressive Moving Average) 모델 파라미터들은 블록(705)에서, 신호들(x₁[n] 및 x2[n]) 사이의 매그니튜드 응답 차이에 대응하도록 계산된다. 그 후, ARMA 모델 파라미터들은 블록(706)에서, 입력 신호들(x₁[n] 및 x2[n]) 사이의 매그니튜드 응답 차이에 대응하는 시간 도메인 임펄스 응답을 추정하는데 사용될 수 있다. 추정된 시간 도메인 임펄스 응답은 선형 위상을 갖는 시간 도메인 임펄스 응답 필터를 생성하도록 제한될 수 있다. 그 후, 신호들 중 하나, 예컨대, x2[n]는 블록(708)에서, 블록들(706 및 707)에서 계산된 제한된 시간 도메인 임펄스 응답을 사용하여 필터링된다. 부가적으로, 필터링되지 않은 신호(x₁[n])는 일부 실시예들에서, 블록(708)에서 적용된 시간 도메인 임펄스 응답 필터에 의해 야기된 지연을 보상하도록 지연될 수 있다. [0056] In some embodiments, the magnitude response used in generating the adaptive filter for equalizing signals in the time domain may be calculated using statistical approximations of the time domain representations of the signals. 7 is an exemplary signal processing flow for matching a magnitude response in the time domain in accordance with one embodiment of the present disclosure. Adaptive filters may be used, for example,

May be based on the magnitude response of the signals found using approximations. Figure 7 illustrates an embodiment of input signals (x ₁ [n] and x2 [n]), the blocks 701 and 702 respectively receiving method of the present invention is from. Block 703 computes an estimate of AR (Auto-Regressive) model parameters of signal x ₁ [n] and block 704 computes an estimate of AR model parameters of signal x 2 [n]. Next, the Auto-Regressive Moving Average (ARMA) model parameters are calculated in block 705 to correspond to the magnitude response difference between signals x ₁ [n] and x 2 [n]. The ARMA model parameters may then be used at block 706 to estimate the time domain impulse response corresponding to the magnitude response difference between the input signals x ₁ [n] and x 2 [n]. The estimated time domain impulse response may be limited to produce a time domain impulse response filter with a linear phase. One of the signals, e.g., x2 [n], is then filtered, at block 708, using the limited time domain impulse response computed at blocks 706 and 707. Additionally, the unfiltered signal (x ₁ [n]) may, in some embodiments, be delayed to compensate for the delay caused by the time domain impulse response filter applied at block 708.

[0057] The adaptive filter may be computed using time domain approximations as shown in the system of FIG. The system of FIG. 8 receives input signals x ₁ [n] and x ₂ [n] from the first and second sensors, respectively, at nodes 801 and 802. Nodes 801 and 802 are coupled to respective processing blocks 803 and 804. Processing blocks 803 and 805 calculate LPCs (Linear Prediction Coefficients) for the input signal x ₁ [n] and processing blocks 804 and 806 compute LPC Lt; / RTI > In some embodiments, LPCs are estimated using auto-regressive (AR) model parameters.

근사를 수행한다. 일부 실시예들에서, 프로세싱 블록(805)은 추가로 프로세싱 블록(807)에 커플링되고, 프로세싱 블록(806)은 추가로 프로세싱 블록(808)에 커플링된다. 프로세싱 블록들(807 및 808)은 도 5의 블록들(507-510)과 관련하여 설명된 것과 유사한 평활화를 수행한다. 일부 실시예들에서, 프로세싱 블록들(807 및 808)은 또한, 시간 도메인 임펄스 응답이 선형 위상 응답을 갖도록, 프로세싱 블록들(805 및 806)에서 계산되는 추정된 시간 도메인 임펄스 응답을 제한할 수 있다. 추정된 시간 도메인 임펄스 응답을 제한하는 것은, 예를 들어, 필터링 지연을 적용함으로써 아래에서 설명되는 바와 같이 수행될 수 있다. [0058] Processing block 803 is coupled to processing block 805, and processing block 804 is coupled to processing block 806. Of processing blocks (805 and 806) are input signals (x ₁ [n] and x2 [n]) to receive the LPC and the input signal to the (x ₁ [n] and x2 [n]), each ARMA ( auto-regressive, moving average) Calculates the magnitude response difference between system coefficients. In some embodiments, processing blocks 805 and 806 may then use an ARMA coefficient to approximate a time domain impulse response corresponding to the magnitude difference between the input signals x ₁ [n] and x 2 [n] Using

Perform approximation. In some embodiments, processing block 805 is further coupled to processing block 807, and processing block 806 is further coupled to processing block 808. In some embodiments, Processing blocks 807 and 808 perform a smoothing similar to that described with respect to blocks 507-510 of FIG. In some embodiments, processing blocks 807 and 808 may also limit the estimated time domain impulse response computed in processing blocks 805 and 806 such that the time domain impulse response has a linear phase response . Limiting the estimated time domain impulse response can be performed as described below, for example, by applying a filtering delay.

[0059] Processing block 807 is coupled to error signal processing block 809 where the error signal is computed. Processing block 808 is coupled to adaptive filter 810, where time domain impulse response coefficients are used to generate the adaptive filter. The adaptive filter 810 is further coupled to the error signal processing block 809 via a feedback loop. The adaptive filter 810 generates a filter h [n] to be applied to the original input signal x2 [n] in the processing block 811. [ Some embodiments may include a delay block 812 coupled between the processing block 807 and the error signal processing block 809 to calculate the delay of the time domain impulse response. The computed delay from the delay block 812 is computed from the input signals x ₁ [n] and x ₂ [n] after the x ₂ [n] has been filtered by the adaptive filter h [n] (not shown in FIG. 8) (x ₁ [n]) so as to keep the signal [n] in synchronization.

[0060] For example, in one embodiment, processing blocks 803 and 804 may compute linear prediction coefficients (LPCs) based on the following equation:

here,

이고,

ego,

이다.

to be.

)은 다음의 등식에 기초하여 자동-상관 시퀀스의 추정을 통해 Levinson's-Durbin 알고리즘을 사용하여 추정될 수 있다:The parameters (

) Can be estimated using the Levinson's-Durbin algorithm through the estimation of the auto-correlation sequence based on the following equation:

[0062] In some embodiments, the magnitude response difference computed in processing blocks 805 and 806 may be defined as:

[0063] In some embodiments for calculating the LPC coefficients described above, the adaptive filter may be defined by the following equation:

로 표현되는 이동 평균은 다음의 등식에 정의된 바와 같이 대략적으로 동일할 수 있다:[0064] When processing blocks 805 and 806 apply a Pade approximation, an auto-regressive, moving average system (ARMA), denoted H (z)

Can be roughly the same as defined in the following equation: < RTI ID = 0.0 >

.

.

The approximation can then be expanded and expressed as defined by the following equation:

일 때,

when,

.

.

The coefficients b ₀ ... b _M can then be solved by carrying the denominator on the left to the right and equalizing the coefficients to produce linear systems of equations.

[0065] The coefficients may be limited to a linear phase, for example, by applying a filtering delay. For example, the approximation can then be expanded and expressed by the following equation:

The set of linear equations may be similarly formulated from this equation to equalize the polynomials as discussed above in order to generate a limited set of coefficients to be used in the filter h [n] of the processing blocks 810 and 811 . The linear system of equations can then be solved to obtain the coefficients b ₀ , ..., b _h .

[0066] In some embodiments, the error signal calculated in the error signal processing block 809 may be calculated based on the following equation:

.

.

)이 변하기 때문에, 적응형 인에이블 입력 신호 제어 신호가 참일 때만 업데이트될 수 있다. [0067] In some circumstances, the input signal is composed of noise and speech, and the relative magnitude spectrum of voice and noise can be very different. In such scenarios, always matching the magnitude spectrum can cause undesirable results. Accordingly, some embodiments further include an adaptive enable input signal that controls time instances during which any of the magnitude equalization processing blocks 803-811 are enabled. The adaptive filter h [n] of the processing block 811, only when the magnitude equalization processing blocks are enabled,

Is changed, it can be updated only when the adaptive enable input signal control signal is true.

[0068] The time domain adaptive filters and other components and methods described above may be implemented in an audio controller of a device, such as a mobile device, to process signals received from the near and / or far distance microphones of the mobile device. The mobile device may be, for example, a mobile phone, a tablet computer, a laptop computer, or a wireless earphone. A processor of a mobile device, such as a device's application processor, may implement other processing techniques or processing circuitry as described above with reference to Figures 2, 3, 4, 5, 7, have. Alternatively, the mobile device may include specific hardware for performing these functions, such as a digital signal processor (DSP). The controller may include a processor, a digital signal processor (DSP), and / or other circuitry associated with signal processing. In some embodiments, the controller is coupled to an audio coder / decoder (CODEC) with other audio processing circuitry such as adaptive echo cancellation (AEC), adaptive noise cancellation (ANC), pulse width modulators (PWM) Chip. ≪ / RTI >

[0069] The schematic flow charts of Figures 3, 4, 5, 7, and 8 are generally described as a logical flow chart. Accordingly, the order shown and labeled steps represent aspects of the disclosed method. Other steps and methods that are equivalent in function, logic, or effect to one or more steps or portions thereof of the illustrated method may be contemplated. Additionally, the format and symbols used are provided to illustrate the logical steps of the method, and are understood to not limit the scope of the method. While various arrow types and line types may be used in the flow diagrams, it is understood that they do not limit the scope of the corresponding method. Indeed, some arrows or other connectors may only be used to indicate the logical flow of the method. For example, the arrows may indicate a waiting or monitoring period of time that is not specified between the listed steps of the depicted method. Additionally, the order in which a particular method occurs may or may not strictly adhere to the order of the steps shown.

[0070] When implemented in firmware and / or software, the functions described above may be stored as one or more instructions or code on a computer-readable medium. Examples include non-transitory computer-readable media encoded in a data structure and computer-readable media encoded with a computer program. Computer-readable media include physical computer storage media. The storage medium may be any available media that can be accessed by a computer. By way of example, and not limitation, such computer-readable media can comprise random-access memory (RAM), read-only memory (ROM), electrically-erasable programmable read-only memory (EEPROM) ) Or any other optical disk storage, magnetic disk storage or other magnetic storage devices, or any other medium which can be used to store the desired program code in the form of instructions or data structures, and which can be accessed by a computer . Disks and discs are used to store information such as compact discs (CDs), laser discs, optical discs, digital versatile discs (DVD), floppy discs and Blu- . Generally, discs reproduce data magnetically, and discs reproduce data optically. Combinations of the above should also be included within the scope of computer-readable media.

[0071] In addition to storing on a computer readable medium, instructions and / or data may be provided as signals on transmission media included in a communication device. For example, the communication device may include a transceiver having signals indicative of instructions and data. The instructions and data are configured to cause one or more processors to implement the functions outlined in the claims.

[0072] While this disclosure and certain exemplary advantages have been described in detail, it should be understood that various changes, substitutions and alterations can be made herein without departing from the spirit and scope of the present disclosure as defined by the appended claims It should be understood. Furthermore, the scope of the present application is not intended to be limited to the specific embodiments of the process, machine, manufacture, composition of materials, means, methods and steps described in the specification. As those skilled in the art will readily appreciate from the present disclosure, there are currently existing or later developed processes, which perform substantially the same functions or achieve substantially the same result as the corresponding embodiments described herein, , Manufacture, composition of matter, means, methods, or steps may be utilized. Accordingly, the appended claims are intended to include within their scope such processes, machines, manufacture, composition of matter, means, methods, or steps.

Claims (24)

As a method,
Receiving, by a processor coupled to the plurality of sensors, at least a first input signal and a second input signal in the time domain from the plurality of sensors;
Converting, by the processor, the first input signal and the second input signal from the time domain to a first frequency domain input signal and a second frequency domain input signal;
Wherein the processor is configured to generate a first frequency domain input signal and a second frequency domain input signal based at least in part on the first frequency domain input signal and the second frequency domain input signal, wherein only the first time domain magnitude only equivalent magnitude only equivalent signal and the second time domain magnitude are equivalent Estimating a signal;
Using the time domain adaptive filter, by the processor to filter only the first time domain magnitude equivalent signal to match only the second time domain magnitude to an equivalent signal;
The processor updates the coefficients for the impulse response of the adaptive filter so as to minimize the difference between only the first time domain magnitude equivalent signal and only the second time domain magnitude equivalent signal, ;
Limiting the updated coefficients of the adaptive filter by the processor such that the impulse response of the adaptive filter is limited to having a linear phase response; And
And filtering, by the processor, at least one of the first input signal and the second input signal based at least in part on a limited time domain impulse response.
Way. delete The method according to claim 1,
Further comprising repeating the receiving, estimating, converting, limiting, updating and filtering steps to provide adaptive equalization of the received input signals ,
Way. The method according to claim 1,
Wherein the limiting step comprises limiting the filter coefficients to even symmetric and odd lengths, and wherein filtering comprises applying an adaptive filter having filter coefficients that are computed and limited,
Way. The method according to claim 1,
Further comprising delaying at least one of the first input signal and the second input signal that is not filtered based on the limited time domain impulse response to compensate for the delay introduced by the filtering.
Way. The method according to claim 1,
Wherein the first input signal and the filtered second input signal are further filtered for spatial recognition,
Way. The method according to claim 1,
Wherein the first input signal and the filtered second input signal are further filtered for beamforming,
Way. As an apparatus,
A first input node configured to receive a first input signal;
A second input node configured to receive a second input signal;
And a controller coupled to the first input node and coupled to the second input node,
Receiving the first input signal and the second input signal in a time domain;
Converting the first input signal and the second input signal from the time domain into a first frequency domain input signal and a second frequency domain input signal;
Estimating a signal that is equivalent only of a first time domain magnitude and a signal of which only a second time domain magnitude is equivalent based at least in part on the first frequency domain input signal and the second frequency domain input signal;
Using a time domain adaptive filter to filter only the first time domain magnitude equivalent signal to match only the second time domain magnitude to an equivalent signal;
Updating the coefficients for the impulse response of the adaptive filter to minimize a difference between the first time domain magnitude only equivalent signal and the second time domain magnitude only equivalent signal;
Limiting the updated coefficients of the adaptive filter such that the impulse response of the adaptive filter is limited to having a linear phase response; And
And filtering at least one of the first input signal and the second input signal based at least in part on a limited time domain impulse response.
Device. delete 9. The method of claim 8,
Further comprising repeating the receiving, estimating, converting, limiting, updating and filtering steps to provide adaptive equalization of the received input signals ,
Device. 9. The method of claim 8,
Wherein the limiting step comprises limiting the filter coefficients to even symmetric and odd lengths, and wherein filtering comprises applying an adaptive filter having filter coefficients that are computed and limited,
Device. 9. The method of claim 8,
Further comprising delaying at least one of the first input signal and the second input signal that is not filtered based on the limited time domain impulse response to compensate for the delay introduced by the filtering.
Device. 9. The method of claim 8,
Wherein the first input signal and the filtered second input signal are further filtered for spatial recognition,
Device. 9. The method of claim 8,
Wherein the first input signal and the filtered second input signal are further filtered for beamforming,
Device. As a method,
Receiving at least a first input signal and a second input signal in a time domain from a plurality of sensors, by a processor;
Computing, by the processor, auto-regressive (AR) model parameters of the input signals using linear prediction analysis;
Computing, by the processor, auto-regressive moving average (ARMA) model parameters corresponding to the magnitude response differences between the two input signals;
Computing, by the processor, a time domain impulse response corresponding to a magnitude response difference between the first input signal and the second input signal, wherein the time domain impulse response comprises the autoregressive model parameters and the auto- Based at least in part on moving average model parameters

Calculated using approximation;
Limiting, by the processor, the time domain impulse response to have a linear phase response; And
And filtering, by the processor, at least one of the first input signal and the second input signal based at least in part on a limited time domain impulse response.
Way. 16. The method of claim 15,
Wherein applying the linear prediction analysis comprises generating linear prediction coefficients.
Way. 16. The method of claim 15,
Wherein the first input signal and the second input signal comprise audio information received from a first microphone and a second microphone,
Way. 16. The method of claim 15,
Wherein the first input signal and the filtered second input signal are further filtered for spatial recognition,
Way. 16. The method of claim 15,
Wherein the first input signal and the filtered second input signal are further filtered for beamforming,
Way. As an apparatus,
A first input node configured to receive a first audio signal;
A second input node configured to receive a second audio signal;
And a controller coupled to the first input node and coupled to the second input node,
Receiving a first input signal and a second input signal in a time domain;
Computing, by the processor, automatic-regression (AR) model parameters of the input signals using linear prediction analysis;
Computing, by the processor, auto-regressive moving average (ARMA) model parameters corresponding to a magnitude response difference between two input signals;
Computing, by the processor, a time domain impulse response corresponding to a magnitude response difference between the first input signal and the second input signal, wherein the time domain impulse response comprises the autoregressive model parameters and the auto- Based at least in part on moving average model parameters

Calculated using approximation;
Limiting the time domain impulse response to have a linear phase response; And
And filtering at least one of the first input signal and the second input signal based at least in part on a limited time domain impulse response.
Device. 21. The method of claim 20,
The controller is further configured to generate linear prediction coefficients when computing automatic-regression (AR) model parameters of the input signals by using linear prediction analysis.
Device. 21. The method of claim 20,
Wherein the first input signal and the second input signal comprise audio information received from a first microphone and a second microphone,
Device. 21. The method of claim 20,
Wherein the first input signal and the filtered second input signal are further filtered for spatial recognition,
Device. 21. The method of claim 20,
Wherein the first input signal and the filtered second input signal are further filtered for beamforming,
Device.

Images